| 00:59.26 | Notify | 03BRL-CAD:starseeker * 66976 brlcad/trunk/src/libgcv/plugins/obj/wfobj/obj_rules.l: remove commented out header |
| 01:01.10 | Notify | 03BRL-CAD:starseeker * 66977 brlcad/trunk/regress/repository.sh: Couple more files to ignore |
| 01:18.09 | starseeker | brlcad: not sure why we haven't seen it before, but I'm getting a benchmark failure on the odd pathnames regression test. |
| 01:18.29 | starseeker | brlcad: it looks like the "look_for" function is having problems when fed paths with spaces in them |
| 01:18.54 | starseeker | so far I haven't been able to figure out what the necessary quoting is |
| 01:19.42 | starseeker | (in the run.sh file, or bin/benchmark when copied to the build directory) |
| 01:22.13 | starseeker | brlcad: conceptually, could benchmark be a C++ compiled program? |
| 01:22.51 | starseeker | is trying to figure out how to do what that does portably... sh is out for Windows reasons, CMake is a bad idea because we want benchmark to run from an installed BRL-CAD even without CMake on the system... |
| 02:18.40 | *** join/#brlcad pefdheytrnbwhnmb (~armin@dslb-088-066-141-131.088.066.pools.vodafone-ip.de) | |
| 04:14.19 | andromeda-galaxy | brlcad: when you decide on what you think is the best way around the repository issue, ping me & mention it, I'll see it sometime |
| 04:14.46 | andromeda-galaxy | starseeker: we can't just bundle mingw bash?H |
| 04:59.38 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 05:25.22 | brlcad | starseeker: think I found the problem |
| 05:25.49 | brlcad | rather, definitely found at least one problem that is probably why odd pathnames is failing |
| 05:26.17 | brlcad | the benchmark itself and look_for is handling spaces and other odd chars fine |
| 05:29.39 | brlcad | it's the command-line argument handling, not quoting the args or the cwd for automatic searching |
| 06:11.25 | Notify | 03BRL-CAD:brlcad * 66978 brlcad/trunk/bench/run.sh: change to a while loop so argument override handling will properly handle variables with spaces (e.g., RT=/path\ to/foo/bin/rt). make is more clear that any argsremaining are assumed to be rt args. |
| 06:12.51 | brlcad | starseeker: and way ahead of you there on the portability thoughts |
| 06:15.00 | brlcad | andromeda-galaxy: so it's sounding like git annex workflow isn't really going to work anyways -- what about using github LFS? |
| 06:15.23 | brlcad | I'd assumed annex would only refer to canonical path on .bz |
| 06:15.36 | brlcad | not each author's clone |
| 08:29.14 | Notify | 03BRL-CAD:brlcad * 66979 brlcad/trunk/bench/run.sh: space handling issues were a bit more complex than initially diagnosed. needed to quote cwd paths being searched and eval quoted command invocations. untested handling of PATH with spaces, and need to make sure eval is behaving on some non-bsd cli variants. |
| 10:34.37 | *** join/#brlcad merzo (~merzo@user-94-45-58-141.skif.com.ua) | |
| 11:28.23 | Notify | 03BRL-CAD:brlcad * 66980 brlcad/trunk/misc/CMake/distcheck_repo_verify.cmake.in: some distcheck fixes; wasn't working from a source/build path containing spaces |
| 12:03.45 | ``Erik | curl and wget are nice and all, but fetch is where it's at, yo! |
| 12:43.44 | starseeker | brlcad: thanks for the catch (r66980) - should have spotted that long since |
| 12:44.10 | starseeker | shakes head at self... no wonder I was seeing that test finish earlier in the logs |
| 14:45.29 | Notify | 03BRL-CAD:starseeker * 66981 brlcad/trunk/misc/CMake/distcheck_repo_verify.cmake.in: Don't want quotes around list of paths when doing list processing operations. |
| 15:22.44 | andromeda-galaxy | brlcad: I'm taking a look at lfs now, let me make sure it doesn't have the same problems.. |
| 15:31.13 | starseeker | brlcad: odd pathnames benchmark is still failing |
| 15:31.21 | starseeker | can't find rt |
| 15:32.29 | starseeker | what's the result of the thinking for portable benchmark? half tempted to see if I can throw a C++ program together, if that's the answer... |
| 15:41.56 | andromeda-galaxy | brlcad: it looks to me like git lfs expects some special kind of server of its own, it won't track files that are just available over http, so we'd have to stick the files in github. I'm not clear on whether they're git lfs implementation is okay with 3GB of files per repo, though |
| 15:56.57 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 16:06.15 | Notify | 03BRL-CAD:d_rossberg * 66982 brlcad/trunk/src/conv/patch/rpatch.c: a small change which makes MS Windows users happy: omit the carriage returns at lines' ends which confuse patch-g |
| 16:53.12 | starseeker | andromeda-galaxy: even if it is, I'd be surprised if github would go for that... |
| 17:02.51 | Notify | 03BRL-CAD:starseeker * 66983 (brlcad/trunk/CMakeLists.txt brlcad/trunk/misc/doxygen/CMakeLists.txt brlcad/trunk/src/CMakeLists.txt): Proof-of-concept BRLCAD_FEATURES macro that produces a Features list in the doxygen output. |
| 17:03.28 | starseeker | brlcad: 66983 is the "flat" feature list - essentially entries with an optional description. |
| 17:03.33 | andromeda-galaxy | starseeker: indeed. they have a thing about 1G repos somewhere else, but that page says something about how you should version stuff with git lfs if its too big, so I'm not sure what their limits are |
| 17:59.45 | *** join/#brlcad Ch3ck (uid129553@gateway/web/irccloud.com/x-ckjltauhrdnkcpwu) | |
| 18:37.20 | *** join/#brlcad JulytoNovember (~ubuntu@2602:306:3b33:9ac0:6c76:2406:5159:478e) | |
| 18:59.18 | Notify | 03BRL-CAD:starseeker * 66984 brlcad/trunk/include/vmath.h: Tweaks to vmath.h doxygen. |
| 19:52.42 | *** join/#brlcad merzo (~merzo@185-47-133-95.pool.ukrtel.net) | |
| 21:00.56 | brlcad | andromeda-galaxy: my understanding was that lfs was specifically so you can go over (WAY over) with lfs because that's what it's designed for |
| 21:01.59 | brlcad | starseeker: way past thinking ... don't worry about it? :) |
| 21:02.33 | brlcad | I wasn't done with benchmark odd pathnames -- distcheck-full is still giving me some fits |
| 21:09.09 | starseeker | brlcad: fair enough |
| 21:10.56 | *** join/#brlcad merzo (~merzo@26-28-133-95.pool.ukrtel.net) | |
| 21:15.59 | andromeda-galaxy | brlcad: indeed... so you think that using lfs & putting all of the data on github might be better than hosting the data ourselvs? |
| 21:41.48 | brlcad | andromeda-galaxy: yes if there's value in having it in a repo, collaborative value ... and I think there probably is for this for the reasons we talked about yesterday |
| 21:42.42 | brlcad | the only problem I just found is that lfs is limited to 1GB |
| 21:42.50 | brlcad | (we could pay for more) |
| 23:06.24 | andromeda-galaxy | brlcad: that's a problem (and part of what I was worrying about earlier)... do you know of any other decent alternatives? |
| 23:16.55 | brlcad | andromeda-galaxy: let me think on it... |
| 23:45.01 | *** join/#brlcad merzo (~merzo@126-245-134-95.pool.ukrtel.net) | |